back to article Smartphone SatNavs to get centimetre-perfect GNSS receivers in 2018

Broadcom is now sampling silicon it says will make smartphones' SatNav systems accurate to a centimeter in 2018. The company announced the new "BCM47755" chipset at the ION GNSS+ conference in Portland, Oregon, and says it will deliver 30cm accuracy even in the concrete canyons of cities where today's GPS can struggle. The …

  1. Lee D Silver badge

    That's all well and good.

    But does the mapping, and indeed the routing software, actually care about that level of accuracy.

    Take driving off a motorway intersection. My sat-nav is always convinced I've continued onwards until, quite literally, I'm off the motorway by some tens of meters. The software cheats and just assumes you've followed course until you hit a big error margin at which point it's willing to accept defeat and recalculate.

    How are you going to utilise cm-level accuracy without a) bugging the user because he drifted slightly left or b) having to "debounce" all the location logic so it doesn't make decisions based on tiny variations?

    "Better signal" in difficult locations is great. But the accuracy thing is a bad sell.

    1. The Mole

      The reason your sat-nav does that is because at speed the accuracy of gps can easily be +-50m and so it simply doesn't know reliably whether you actually have left the motorway or not. If the accuracy is reduced to +-5m at speed (>1m level when stopped) then it can much more reliably tell the difference between being on the slip road compared to being on the motorway as the hard shoulder/verge now provides sufficient separation of the error radius.

      1. Charles 9 Silver badge

        How accurate is this new chipset supposed to be at motorway speeds? I would think it needs to be accurate to less than 2m in order to be able to distinguish specific lanes.

        1. Dave 126 Silver badge

          > How accurate is this new chipset supposed to be at motorway speeds?I would think it needs to be accurate to less than 2m in order to be able to distinguish specific lanes.

          I for one travel at motorway speeds within a lane and not across them! Phone GPS units usually refresh once per second, some automotive units at 5 or 10 times a second. The accuracy of GPS isn't dented by speed (which is also calculated from the signals).

  2. Lysenko

    Well, that'll be useful for giving your Molotov cocktail equipped drone an autonomous "precision strike" capability.

    Hmmm ... that can't be it. No $$$ in that. What are they thinking of? Aha! Parking fees? If you can pin the vehicle to a precise parking spot you can charge more optimise efficiency and improve customer service. Also useful for pinning down exactly which hotel room your wife is meeting her spiritual adviser/personal trainer/tennis coach in.

    1. Charlie Clark Silver badge

      Industry and agriculture are driving the demand for this kind of precision.

      1. Adam 1 Silver badge

        but not just that

        Airports are big places. It can be tricky to understand which taxiway the fruity map app is directing me to drive down.

  3. Anonymous Coward
    Anonymous Coward

    There are presumably some road systems where there are vertical tiers running in the same direction. Would the GNSS systems determine which one you were on by calculating your height too? Would the height difference between a sports car and a truck cab make a difference?

    1. Charles 9 Silver badge

      GPS and the like have always been three-dimensional (lat, long, and alt). Though I've never driven a double-decker roadway lately so I can't be sure how well they work. And not just the same direction, what about double-deckers that run in opposite directions? My system already has difficulty with reversible HOV lanes.

      1. Dave 126 Silver badge

        Differentiating between lanes separated only by height is an issue for the map data - the GPS unit itself can give you a 3-dimensional location coordinate. If the map data is in 3D - or more likely a format derived from past traffic GPS data, eg the upper lane eventually veers East) - there's no problem.

    2. ciaran

      GPS is bad at measuring height

      Basically no, the height part of the GPS measurement has a much larger margin of error. The lateral measurements are best when the satellites are low and well separated, while the height measurement will be better when the satellites are basically directly above you.

      Also you want the height over the ground, but the GPS maps don't have accurate surface height information. The GPS measurement gives your location on a sphere, which the earth isn't.

      1. David Harper 1

        Re: GPS is bad at measuring height

        "The GPS measurement gives your location on a sphere, which the earth isn't."

        No it doesn't. The default GPS reference geoid is the WGS84 ellipsoid, which is an oblate spheroid that closely matches global mean sea level. If your GPS receiver can pick up signals from more than three satellites, then it can calculate a full 3-D position consisting of latitude, longitude and height above the reference geoid. There are enough GPS satellites these days that a 3-D fix with good height-above-geoid accuracy is pretty easy to obtain, even with a mobile phone.

        1. Anomyous Curd

          Re: GPS is bad at measuring height

          Woah there. Just because GPS returns a number for "height above geoid" doesn't mean you should use it, for two main reasons:

          - GPS vertical accuracy is inherently lower, as satellite positioning is optimised for "horizontal" accuracy and the two are exclusive. This is exacerbated on cheap, low-power receivers that only process the minimum of information needed.

          - The WGS84/EGM96 geoid is hideously inaccurate by today's standards, the current (2004) iteration of the ellipsoid provides resolution to something on the order of 100km. So the value you're getting is only telling you how high you are above the geoid's height averaged over the surrounding hundred kilometres.

          Now, this can be corrected, but you're typically talking an assisted GPS solution based on some known reference altitude, using the GPS signals only to correct for motion relative to that point.

          Correcting for position in 3d road systems does not involve altitude values at all. Remember roads are basically graphs. You know which ways you can come in and come out (from the map) and you know which way you did come in (from the "horizontal" data) so you know which layer you ended up on.

          1. Charles 9 Silver badge

            Re: GPS is bad at measuring height

            "Correcting for position in 3d road systems does not involve altitude values at all. Remember roads are basically graphs. You know which ways you can come in and come out (from the map) and you know which way you did come in (from the "horizontal" data) so you know which layer you ended up on."

            Which is useless if it's turned on en route or otherwise loses context along the way so it doesn't know from which way you came. Plus what if BOTH levels can be reached from the same route (very possible--George Washington Bridge and Verazanno-Narrows Bridge in NYC are both bidirectional double-decker bridges with both decks accessible from both ends).

            1. Anomynous Curd

              Re: GPS is bad at measuring height

              In which case, respectively:

              1) If you lose the signal you lose the signal, what's the difference whether it's up, down, left, right?

              2) No. Short of a bridge with vertical a car elevator (which would be freakin' awesome) you're still coming in on a physically different road, which is easily distinguished with sufficiently precise data, which this is.

              1. Charles 9 Silver badge

                Re: GPS is bad at measuring height

                "If you lose the signal you lose the signal, what's the difference whether it's up, down, left, right?"

                Because it could lose the signals just as you get on. End result, it doesn't know which deck you entered. Murphy's Law. Like I said, most satnavs I've tried (including Google Maps) get mixed up simply by going into a reversible lane.

                "No. Short of a bridge with vertical a car elevator (which would be freakin' awesome) you're still coming in on a physically different road, which is easily distinguished with sufficiently precise data, which this is."

                I still haven't gotten a clear answer to just HOW precise the thing is at motorway speeds. It needs to be within 2m while driving 100km/h to not get mixed up, especially if the ramps involved are very close to each other (which actually happens at those two bridges; I used to live on Long Island). Other complicated interchanges like Bruckner and Washington's Springfield Interchange create similar mixups.

      2. Jan 0

        Re: GPS is bad at measuring height

        I'm aware that some bicycle computers and satnavs have incorporated barometers for the last 20 years or so. They're obviously as crap as GPS for giving your true altitude, but they can measure your rate of climb or descent. A phone or vehicle satnav could also incorporate a barometer and calculate which tier of a road you are on.

      3. Anonymous Coward
        Anonymous Coward

        Re: GPS is bad at measuring height

        My GPS is quite good at measuring height, and does so to a about 1 metre reporting accuracy. Sometimes when I need amusement, I switch on height display, and mentally map the altitudes of various local roads and locations.

        GPS measurements are not based on a sphere, but on a 'geodetic datum' which defines a model shape for the earth which is a better approximation than a sphere.

        The exact datum used varies, depending on whether more priority is being given to universal (all over the world) performance, or local geodetic systems intended to be more accurate over a smaller area.

    3. Anonymous Coward
      Anonymous Coward

      Vertical tiers?

      "Would the GNSS systems determine which one you were on by calculating your height too? "

      If you're on the lower tier I'd expect GPS signal quality would be greatly reduced (or non-existent) so my "smart nav" would know, no GPS - I must be on the lower tier. As to where I am on that lower tier, sorry, please try again later.

  4. James 51 Silver badge

    Any chance of combining the various GPS systems for more accuracy?

    1. Charles 9 Silver badge

      They're generally compatible with each other already. They run on similar principles. You just gotta know the frequencies involved. Modern satnav chips IINM can tap into any and all that are available.

    2. S4qFBxkFFg

      It should. AFAIK, if they don't support GLONASS they are prohibited from using it in phones sold in Russia, which is a large enough market to matter. I imagine the same is effectively true in China with Beidou, even if not formalised.

      1. Knoydart

        Re: Never do this

        My entry level Garmin can choose which constellation (GPS or GLONASS) it uses but doesn’t seem to be able to combine them to increase accuracy.

        There is work for aeronautical use on having duplicate constellations being able to be received by receivers to provide diversity of reception from different constellations but I get that GPS at the moment is the only one the American's trust for its civil aviation fleet and so on. ICAO will have a big job on their hands qualifying many of the GNSS constellations for worldwide use.

        1. Andrew Moore Silver badge

          Re: Never do this

          My kit's currently using GPS, GLONASS, Galileo and Beidou all simultaneously (QZSS is also possible but no chance we'll see that in the northern hemisphere).

    3. This post has been deleted by its author

  5. Ralph the Wonder Llama
    Joke

    "...will let SatNav systems advise which lane you should drive in"

    I can see it now:

    If {car.Marque == marques.BMW || car.Marque == marques.Mercedes || car.Marque == marques.Audi}

    {

    lane = lanes.Outside;

    }

    1. Lysenko

      Re: "...will let SatNav systems advise which lane you should drive in"

      Careful with that fuzzy equality operator. "car.Marque == marques.Mercedes" might catch some Chryslers. Always use "===" when you need to set "driver.Tosser = True".

    2. Warm Braw Silver badge

      Re: "...will let SatNav systems advise which lane you should drive in"

      The 30cm accuracy is probably required to detect the separation between the Audi in the outside lane and the car in front of it...

    3. JamesPond
      Joke

      Re: "...will let SatNav systems advise which lane you should drive in"

      ELSE IF Lane.outside = Full THEN DoUndertake

  6. Knoydart
    Boffin

    What's the frequecny Kenneth?

    I think you'll find that L1 is up at 1575MHz and L5 is down at 1176MHz. Therefore the high / low frequency mix you mention is not correct.

    1. James Wilson
      Happy

      Re: What's the frequecny Kenneth?

      +1 for the REM reference. I'll let you off the typo.

  7. Anonymous Coward
    Anonymous Coward

    I am an Audi driver...

    So I am alway in the correct lane.

    1. Lysenko

      Re: I am an Audi driver...

      Unless you're blocking a Bentley, Bugatti, Lamborghini or Porsche in which case as a vulgar, plebeian (distant) cousin you should get the hell out of the way!

    2. Anonymous Coward
      Anonymous Coward

      Re: I am an Audi driver...

      "I am an Audi driver...

      So I am alway (sic) in the correct lane."

      It would be nice if you could use your indicators to let the rest of us know when the correct lane changes....

      1. Adam 1 Silver badge

        Re: I am an Audi driver...

        Audi blinkers aren't for lane changing. They are rather used to signal that the traffic lane has been temporarily designated as a parking spot.

  8. Len Silver badge

    New iPhone combining GPS, Galileo and GLONASS

    Allegedly the new iPhones support combining GPS, Galileo and GLONASS signals for greater accuracy. https://www.macrumors.com/2017/09/26/iphone-8-iphone-x-galileo-satellite-support/

    Does anyone know how the greater accuracy with Broadcom's new chip relates to greater accuracy combining the various systems? And what about reliability? Is combining multiple systems more reliable?

  9. Named coward

    which lane

    Don't all GPS units and phone software already have lane assistance? It doesn't need to know which lane you're on to tell you which lane you're supposed to be on

  10. Wade Burchette

    And yet ...

    And yet Google Maps will still annoy you to turn on the "high precision" accuracy of your phone and if you do not, randomly self-gimp itself. On some drives, I had Google maps tell me it was searching for a signal every 2 or 3 miles. I switched to Waze and Open Street Map, the location was always working and accurate. Went back to Google maps, same problem. I still don't want high-precision accuracy enabled on my phone because who knows what Google will do with it.

    1. DropBear Silver badge

      Re: And yet ...

      You do realise "enable high precision location" just means "actually turn on GPS", right? As opposed to just relying on an approximate based on which cell tower you're pinging off? I absolutely can understand the desire to not continuously report to The Google Mothership where you are at all times, but it's kinda implied you'd need to let Google access that GPS if you want to know where you are on their map, beyond the generic "somewhere in this area". It's not like Waze or OSM can work without "high precision location" enabled (="do please turn the GPS on and allow me access or else I can't do squat either"). Incidentally, why is telling Waze where you are at all times better than telling Google?

  11. TJ1

    Aiming for (semi-) autonomous vehicles?

    Sounds like this is the kind of back-up data source that would aid many quasi-autonomous driver assistance systems.

    If it is accurate enough to differentiate lanes, or even possibly lane-drift, it could act as a component of the position awareness/warning system. Think of drivers allowing the vehicle to drift due to being distracted, dozing, arguing with kids in the back, etc.

    If we're heading for a world of 'connected' vehicles (in the sense of each transmitting its position and velocity to the immediate surroundings) it also offers options to prevent driving too close to other vehicles even in conditions where LIDAR, cameras, and other sensors become unreliable.

  12. theModge

    Very exciting for the railways

    Funky modern signalling systems that actually know where the train is have issues with not knowing which of several parallel lines a train is on, purely from GPS, so there is a need to use Balises on the track for location. If this could be made reliable enough then in 40 years time when it clears all the committees that will need to discuss it, it could be very useful.

    1. Barrie Shepherd

      Re: Very exciting for the railways

      With the "status quo protecting, ultra cautious" signal engineers of the railway companies, the "lets not share technology" supply industry and the "retirement homes" that are the committees of the UIC (with each successive meeting in another country) I suspect that your 40 years is a conservative estimate :-) And when a solution is agreed it will be so complex, over engineered and expensive that it will take another 40 years to implement by which time something else will be in the 'Committee stage".. :-)

  13. FlossyThePig

    The company thinks that level of accuracy will let SatNav systems advise which lane you should drive in, making it a bit easier to prepare for a turn

    Can it set a "last moment" parameter for the idiots who cut across at least one lane to get to the exit with one or two yards to spare before the junction is passed?

  14. David Nash Silver badge

    Location services

    "Weary smartphone owners who find themselves choosing between location services and battery life privacy"

    Fixed that.

  15. c1ue

    Utter bollocks

    This is an excellent example of garbage marketing.

    Yes, *in theory* use of the different band allows greater precision. But so does sampling more than 3 satellites. Given the GLONASS, Compass, Beidou, and other satellite positioning systems available, the feasibility of using many satellites vs. just 3 GPS ones is actually more feasible/easier.

    More importantly, however, the problem in "concrete canyons" isn't the precision of the distance calculation and/or clock data. The problem is multi-path interference.

    Since the GPS signals are radio, reflections of same signal off the ground, buildings, etc arrive at different times than the straight path signal. If the straight path signal is present, you can calculate the distance to the satellite and discard the longer paths (reflections). However, if the straight line path is not present - i.e. blocked by a tall building or whatnot, you *cannot* distinguish a multipath reflection from the straight line unless you have also know the position of the satellite vs observer plus an accurate 3D map of the buildings.

    I can pretty much guarantee the chip doesn't do that.

    1. Steve Todd

      Re: Utter bollocks

      Receiving a second set of data on a different frequency band solves that one. This chip uses precisely that method to help gain accuracy. GALILEO uses up to three bands, a better encoding scheme and a few other tricks to get down to 1cm resolution.

      1. Charles 9 Silver badge

        Re: Utter bollocks

        Perhaps you can elaborate how multiple frequencies solve the problem of multi-path interference, particularly in the case where the ONLY signals visible are reflected (as in a concrete canyon where direct line-of-sight is blocked by buildings).

  16. Kernel Silver badge

    Not quite

    "The BCM47755 achieves its extra accuracy by locking on to three satellites using the L1-band transmission (the GPS broadcast of satellite location, time, and signature)."

    Actually, any GPS receiver needs a minimum of four satellites to lock on to, otherwise it has no way of determining how inaccurate its internal clock is and making appropriate corrections.

    Four satellites allows the calculation of three potential positions - since you're obviously not actually in three places at once, adjusting the clock to minimize the difference between the three calculated positions provides the feedback loop to ensure the cheap internal clock chip is running as close as possible to the rate of the very expensive (and comparatively large) Cesium clocks in the satellites.

    1. Steve Todd

      Re: Not quite

      No, GPS can get away with only 3 satellites for a fix, but can’t give you altitude with just the 3. Four or more are needed for a full 3D fix.

      http://www.gpsnuts.com/mygps/gps/technical/ed.htm

      1. Kernel Silver badge

        Re: Not quite

        To quote from the somewhat light-weight website you included the URL of:

        "there are four unknowns in the pseudo-range equations: the three components of the receiver’s position, and the offset in the receiver’s clock. To determine its location it then must solve the four equations for the four unknowns."

        and

        "If the receiver “knows” one of the three position components, then measurements from only three satellites are required to solve for the other two components and the receiver clock offset (3 equations and 3 unknowns). If, for example, the altitude is known perfectly, this is equivalent to tracking a satellite directly above (or below) the user, and having a perfect pseudo-range measurement to it. Since a real fourth satellite will never provide a perfect pseudo-range, the three-satellite solution will be better than the four-satellite solution (assuming similar geometry quality, i.e., PDOP). If on the other hand the assumed altitude is in error, then the two sides of the equations will be equal only if a compensating error is added to the horizontal position components and clock offset. "

        or. to put it another way, the receiver can only calculate an accurate position from three satellites if it accurately knows its position or its clock offset already - and if it already knows (ie., has been told) its position accurately, then why bother at all?

  17. GcdJ

    I seem to recall the GPS sats have a dither error parttern

    I seem to recall that the GPS sats enforce a randomly changing dither pattern for all non-military uses.

    Receivers can then correct for the differ if a know position transmits its dither delta. int he UK this is done using the BBC radio masts

    Are we saying the dither has now been disabled - or the correction is now so widespread that we just ignore it in conversation.

    I recall that network rail in the UK is very keen to have accurate satellite positioning so that it can find its track side junction boxes. It appears that Network Rail keeps loosing them and then goes to the cost of sending out a search parties before the maintenance can take place..

    1. Charles 9 Silver badge

      Re: I seem to recall the GPS sats have a dither error parttern

      I think they either turned the dither off or reduced its range after consumer tech caught up and was able to compensate for the dither, rendering the practice moot.

    2. Steve Todd

      Re: I seem to recall the GPS sats have a dither error parttern

      You're thinking of the SA (selective availability) code. This was a deliberate random wobble introduced into the civilian (C/A) signal, while military receivers used the more accurate P signal. SA was disabled in 2000 by presidential order, and the latest block III satellites aren't capable of transmitting it.

      1. Charles 9 Silver badge

        Re: I seem to recall the GPS sats have a dither error parttern

        And as I recall, the EO was due to it being moot due to things like A-GPS and other commercially-available means to correct out the wobble. Basically, if someone wanted REALLY accurate information, they could do it whether SA was there or not, so remove it and let the higher accuracy be put to use commercially. Most drift these days is due to physics (atmospheric interference and so on), not sabotage.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2019